HI
it’s normal that “TextDataLoaders.from_df” run on multicores instead on gpu ??
I also specified device arg to: “torch.device(‘cuda’)”; but nothing changes;
thanks
Preprocessing is CPU-bound, yes (IE vocab generation, etc)
a ok, thanks!
HI
it’s normal that “TextDataLoaders.from_df” run on multicores instead on gpu ??
I also specified device arg to: “torch.device(‘cuda’)”; but nothing changes;
thanks
Preprocessing is CPU-bound, yes (IE vocab generation, etc)
a ok, thanks!